T. Le Blevec
Earth Science and Engineering, Imperial College, London, United Kingdom
E-mail: t.le-blevec15@imperial.ac.uk
+447 466 892 985
Royal School of Mines, Prince Consort Road, London SW7 2BP, United Kingdom
O. Dubrule
Earth Science and Engineering, Imperial College and Total, London, United Kingdom
C. John
Earth Science and Engineering, Imperial College, London, United Kingdom
G. Hampson
Earth Science and Engineering, Imperial College, London, United Kingdom
2 Thomas Le Blévec et al.
1 Introduction
Fig. 1 Synthetic examples of facies sequences: (a) non cyclic non rhythmic, (b) non cyclic
but red facies presents rhythmicity, (c) cyclic non rhythmic (d) cyclic rhythmic
This aspect is also shown in Figs. 1b (red facies) and 1d in which thicknesses
between different beds of the same facies are constant.
Rhythmicity can be quantified by geostatistical tools, such as the variogram
and the transiogram which show oscillations or dampened oscillations called
hole effects (Pyrcz and Deutsch, 2014). By looking at the probability density
function (pdf) of facies thicknesses, Ma and Jones (2001) show that as the
coefficient of variation of this pdf decreases, the hole effect becomes more
pronounced. This observation is in agreement with the above remark that
rhythmicity is associated with low thickness variability of vertically stacked
facies cycles. This also explains why Markov chains cannot create hole-effect
transiogram models (Dubrule, 2017), as the corresponding thickness pdf is
exponential (coefficient of variation equal to one). On the other hand, renewal
processes may be able to create dampened hole-effect transiograms, because
they offer the possibility to choose a thickness pdf with a lower coefficient of
variation (Matheron, 1968). However, the transiograms derived from renewal
processes are not always known analytically, and are thus difficult to adapt
well to the observed rhythmicity.
Truncated or (pluri-) Gaussian methods have also been successfully used
to create facies models (Armstrong et al, 2011), and they have been applied
to shallow-marine carbonate reservoirs and outcrop analogues (Amour et al,
2012; Doligez et al, 2011; Le Blévec et al, 2017). The contacts between facies
are defined by the truncation rule applied to a random Gaussian function,
which provides control on facies juxtapositions. However, in its traditional
form the method does not incorporate cyclicity and rhythmicity. Le Blévec
et al (2017) have extended the Pluri-Gaussian method to the modelling of
facies asymmetry in vertical successions, thus creating cyclicity. They produce
asymmetric transition probabilities between facies by introducing a shift in the
correlation of two random Gaussian functions as suggested by Armstrong et al
(2011). This is similar to the approaches of Langlais et al (2008); Renard and
4 Thomas Le Blévec et al.
Beucher (2012), but with more flexibility in the resulting facies transiograms.
Pluri-Gaussian simulations also enable the use of hole effect models (Beucher
and Renard, 2016) and may lead to hole effect facies transiograms that can
generate rhythmicity.
Although cyclicity and rhythmicity are common features of vertical fa-
cies successions, they may have a variable expression laterally, depending on
the formative depositional processes and controls. Laterally extensive facies
in shallow-marine carbonate strata are generally attributed to external (allo-
genic) controls that operated over an entire carbonate platform or shelf, such
as relative sea-level variations (Goldhammer et al, 1990). Facies of limited lat-
eral extent may be attributed to the nucleation, vertical build-up and lateral
shifting of tidal flat islands across a carbonate platform or shelf (Pratt and
James, 1986). This mechanism is internal to the dynamics of the carbonate
platform depositional system (autogenic) and may generate both vertical and
horizontal asymmetry in the stacking of facies if the tidal-flat-island deposits
obey Walthers Law (Burgess et al, 2001; Le Blévec et al, 2016). Cyclic and
rhythmic facies successions can also be overprinted by diagenetic facies after
deposition; for example, hydrothermal dolomite bodies associated with faults
and igneous intrusions are observed to cut across shallow-marine carbonate
platform deposits characterized by rhythmic facies cycles (Jacquemyn et al,
2015).
In order to model two-dimensional areas and three-dimensional volumes
that exhibit cyclic and rhythmic vertical facies successions but different lateral
facies patterns, it is necessary to use different vertical and lateral covariance
models. This is possible via the use of separable anisotropic models (Chiles
and Delfiner, 2012), although such models have rarely been used for facies
modelling (Matheron et al, 1988).
The aim of this paper is to develop a general method based on Pluri-
Gaussian simulations to model facies cyclicity and rhythmicity in the vertical
direction, and a range of appropriate lateral facies patterns using space-time
(lateral-vertical) separable covariance models. After presenting the main defi-
nitions, the three key aspects of the modelling method and their impact on the
transiograms are presented: cyclicity, rhythmicity, separability. A method for
simulating the resulting complex facies architectures is then presented, firstly
for unconditional simulations and then for simulations conditioned to data.
Finally, the method is applied to a case study from the Triassic Latemar car-
bonate platform (northern Italy), which has been widely interpreted to show
cyclicity and rhythmicity.
2 Definitions
The random function representing a facies is the indicator function I(x). If the
facies i is present at a spatial location x, Ii (x) = 1 and if not, Ii (x) = 0. In
Geostatistical modelling of cyclic and rhythmic facies architectures 5
from which can be derived the transiogram tij (h), which is the probability to
transition from facies i to facies j in a certain direction h
Cij (h)
tij (h) = P r{Ij (x + h) = 1|Ii (x) = 1} = . (3)
pi
This transiogram can be asymmetrical, which means that it is different in
opposite direction
tij (h) 6= tij (−h), (4)
and can thus quantify asymmetrical facies successions (Carle and Fogg, 1996;
Le Blévec et al, 2017). The transiograms between different facies are usually
gathered in a transiogram matrix that the simulation method aims to repro-
duce. For instance, a transition matrix between three facies 1, 2, 3 is
t11 (h) t12 (h) t13 (h)
t(h) = t21 (h) t22 (h) t23 (h) . (5)
t31 (h) t32 (h) t33 (h)
The terms on the diagonal are the auto-transiograms and those off-diagonal
are the cross-transiograms. The tangent at the origin of the auto-transiograms
Tii is related to the mean length of the facies i along this direction (Carle and
Fogg, 1996). The tangent at the origin of the cross-transiogram is called the
transition rate
Tij = t0ij (0). (6)
Transition rates are interesting when studying asymmetry because they are
related to juxtaposition between different facies. According to Carle and Fogg
(1996), if a facies j tends to overlie a facies i, rather than being overlaid by it
then
pj
Tij > Tji . (7)
pi
Transition rates are also related to embedded transition probabilities rij (given
the presence of a facies i, the probability that it is overlaid by j)
−Tij
rij = , (8)
Tii
6 Thomas Le Blévec et al.
with g(x) the standardized Gaussian pdf. The transition probability (second
order moment) between facies i and j is
Z q Z q
1
tij (h) = gρ(h) (x, y) dx dy, (13)
pi −∞ −∞
where gρ(h) a standardized bi-Gaussian probability density with correlation
matrix defined by the covariance ρ(h). For q = 0 the two facies have same pro-
portion 1/2 and the analytical solution of this bi-Gaussian integral (Lantuéjoul,
2013; Le Blévec et al, 2017) gives the following auto-transiogram for the two
facies
1 1
t11 (h) = t22 (h) = + arcsin[ρ(h)], (14)
2 π
Geostatistical modelling of cyclic and rhythmic facies architectures 7
pi Facies i proportion
Cij (h) Non centered covariance between facies i and j
tij (h) Transiogram from facies i to facies j
Tij Transition rate from facies i to facies j
rij Embedded transition rate from facies i to facies j
ρ(h) Covariance of the latent standardized bi-Gaussian field Z(x)
q1 , q2 Threshold of the Gaussian field Z1 (x) and Z2 (x)
gρ(h) Standardized bi-Gaussian density with correlation ρ(h)
In other words, every facies transitions into only one other facies upwards
(rij = 1). However, in practice such perfect facies successions are rare and
the embedded transition rates generally do not equal one or zero but have
intermediate values. For instance the sequences of Figs. 1a and 1b show two
identical cycles while the other facies transitions are different. Such sequences
can be called pseudo-cyclic. The limit behaviour, when the sequence is acyclic,
happens if facies have equal probability of transitioning with the other facies,
which would give for three facies the embedded transition rate matrix
0 0.5 0.5
R = 0.5 0 0.5 . (18)
0.5 0.5 0
One can quantify cyclicity with the probability Pc to observe a given cycle
above a given facies, which can be written in the case of three facies for the
cycle composed of facies 1-2-3-1 (under the Markov assumption of indepen-
dence)
Pc = r12 r23 r31 . (19)
We see that the notion of cyclicity is related to asymmetry by using the closing
relations of transiogram matrices
T11 p1 − r21 T22 p2
Pc = r12 (1 − r21 ) , (20)
T33 p3
as the probability Pc increases if the difference between r21 and r12 increases.
Therefore, for a sequence with three facies, asymmetry between two facies re-
sults in cyclicity. The cyclicity studied here (Figs. 1c and 1d) is such that facies
appear only once per cycle. This is incompatible with symmetric cycles that
are not considered here. It also means that every facies tends to transition to
one facies in the upward direction (Eq. (17)) and another one in the downward
direction (which is why asymmetry plays an important role).
Rhythmicity is defined by the repetition of a facies at constant inter-
vals. This is usually observed by computing variograms or transiograms on
sequences that show dampened hole-effects (Journel and Froidevaux, 1982;
Johnson and Dreiss, 1989; Ma et al, 2009). Rhythmicity cannot be quantified
by embedded transition rates as they are independent of facies thicknesses.
Rhythmicity can first be understood when studying two facies. If those fa-
cies have constant thicknesses, the auto-transiogram increases and decreases
rhythmically with a wavelength that is equal to the sum of the two facies
thicknesses (Jones and Ma, 2001). This is similar with more facies as we can
still regard this as the succession of two facies, the one of interest and the one
defined as everything that is not the facies of interest. This is interesting to
consider in combination with cyclicity because the resulting sequences show a
constant cycle thickness (Fig. 1d). Thicknesses along sequences are not usually
constant but can show low variability which results in pseudo-rhythmicity and
dampened hole-effects. The method developed here models pseudo-cyclic and
pseudo-rhythmic sequences, quantified by transiograms.
Geostatistical modelling of cyclic and rhythmic facies architectures 9
Fig. 2 Truncation rule used for the cyclical pluri-Gaussian simulation. q1 and q2 are the
thresholds of the Gaussian functions Z1 (x) and Z2 (x)
Fig. 3 One realization (e) with corresponding transiogram matrix model (a-d) and Gaussian
functions Z1 (x) and Z2 (x) (f). Parameters: ρ = −0.8, r1 = r2 = 0.1, α = 0.05, p1 = p2 = p3
transiograms are asymmetric as in Eqs. (4) and (7). A sensitivity study is also
carried out on the parameters α and ρ showing that the asymmetry can be
controlled by varying these two parameters. Here, the covariance models ρ1 (h)
and ρ2 (h) used for the Gaussian functions Y1 (x) and Y2 (x) are Gaussian
h2
ρ1 (h) = exp(− ), (22a)
r12
h2
ρ2 (h) = exp(− ), (22b)
r22
with parameters r1 and r2 the ranges of the models. An example of a verti-
cal sequence generated with this method and the corresponding facies tran-
siograms is shown in Fig. 3. The simulation method of the latent Gaussian
functions is given in Sect. 4 and the transiograms are computed numerically
(Genz, 1992) based on Eq. (13). Only four transiograms instead of nine are
shown for the three facies (Figs. 3a, 3b, 3c and 3d), because the third facies can
be considered as the background and its transiograms can be directly deduced
from the transiograms of the other facies.
It is clear from Fig. 3 in the realization that the facies are statistically
organized in shallowing upward cycles as highlighted by the low tangent at
the origin of the cross transiogram from intertidal (orange) facies to subti-
dal (red) facies upwards (Fig. 3c). Indeed, this transition is absent along the
vertical section while the opposite transition (from subtidal facies to inter-
tidal facies) occurs 6 times. However, the succession is not completely cyclic
Geostatistical modelling of cyclic and rhythmic facies architectures 11
Fig. 4 Hole-effect on the latent Gaussian field (yellow, Eq. (23)) and the resulting hole-effect
on the facies transition probability (blue, Eq. (24)), with r1 = 0.3 and b1 = 20.
(Fig. 3e) because of a low but non-null probability of the subtidal (red) facies
to transition upwards directly to the supratidal (white) facies. A high variation
of facies thicknesses is also noted, resulting in a non-rhythmic sequence.
Using covariance functions with rhythmicity for the latent Gaussian field is
expected to produce facies transiograms with rhythmicities. This hypothesis
is verified in this section in which a new indicator transiogram model repre-
senting facies rhythmicity is defined. In one dimension, the cosine function is a
valid covariance model (Chiles and Delfiner, 2012) and produces by truncation
vertical sequences with a constant cycle thickness. As this is rarely observed,
a model in which the oscillations attenuate with distance, called dampened
hole-effect (Pyrcz and Deutsch, 2014) gives more flexibility in reproducing cy-
cle thicknesses. Ma and Jones (2001) define the Gaussian cosine covariance as
the product of two valid covariance functions
h2
ρ(h) = exp(− ) ∗ cos(b h). (23)
r2
This covariance gives dampened oscillations controlled by the two parameters
r and b. If the parameter r tends to infinity, the model is the cosine function.
According to Eq. (14) the resulting hole-effect transiogram model is
1 1 h2
t11 (h) = + arcsin[exp(− 2 ) ∗ cos(b h)]. (24)
2 π r
As seen in Fig. 4, this transiogram model has a similar wavelength as the
latent covariance model (Eq. (23)) and the attenuation of the oscillations is
12 Thomas Le Blévec et al.
Fig. 5 One realization (e) with transiogram matrix model (a-d) and latent Gaussian func-
tions (f) with parameters: ρ = −0.82, r1 = r2 = 0.6, b1 = 15, b2 = 30, α = 0.04,
p1 = p2 = p3
3.3 Modelling facies distributions in two and three dimensions with separable
anisotropy
Fig. 6 Covariance map of the latent Gaussian function (left, Eq. (26)) and resulting tran-
siogram (right, Eq. (27)) in two dimensions with rz = 0.5, rx = 1 and b = 20
Equation (27) shows that even though the covariance of the latent Gaussian
field is separable, the resulting transiogram is not separable. Figure 6 com-
pares values (hy constant) of the latent Gaussian function covariance and the
resulting transiogram which both show a dampened hole-effect in the vertical
direction and the absence of a hole-effect along stratigraphy. In intermedi-
ate directions, the dampened hole-effect is present but more attenuated. Once
again, the behavior of the transiogram in two or three dimensions is very sim-
ilar to the covariance of the latent variable. This suggests that there is about
as much flexibility in the transiogram model as in the covariance model.
14 Thomas Le Blévec et al.
Fig. 7 Vertical transiogram matrix model (a-d) and three corresponding realizations (e).
ρ = −0.8, r1z = 0.6, r1x = r1y = 0.4, b1 = 15, b2 = 30, r2z = 0.6, r2x = r2y = 0.4, α = 0.04
These transiogram models are valid for two facies in equal proportions but
for a more general case, different examples in three dimensions are shown in
cross sections with their corresponding transiograms in the vertical direction
(Figs. 7 and 8). The transiograms have been computed by numerical integra-
tion (Genz, 1992) of a Gaussian density (Eq. (13)). The parameters for the
two Gaussian functions are respectively r1x , r1y , r1z and r2x , r2y , r2z , and
vertical hole-effect parameters b1 , b2 .
Figure 7 shows the combination of asymmetry, rhythmicity and anisotropy
in facies distributions in both the realizations and the transiograms. The facies
are ordered according to the asymmetry: subtidal, intertidal and supratidal fa-
cies upwards (Fig. 7e). The vertical thickness of these cycles is almost constant,
confirming the rhythmicity. The different realizations can be interpreted as fa-
cies beds that pinch out laterally or split into different cycles and inter-finger.
Fig. 8 Vertical transiogram matrix model (a-d) and three corresponding realizations (e).
ρ = −0.6, r1z = 0.2, r1x = r1y = 0.4, b1 = 0, b2 = 30, r2z = 0.6, r2x = r2y = 0.4, α = 0.1
of the geometries generated by this transiogram model (Fig. 8), a three di-
mensional simulation with the same transiograms is also presented in Fig. 9.
The orange and white facies clearly show vertical rhythmicity with low lateral
variability in thickness, consistent with a depositional architecture, while the
red (diagenetic) facies truncates them.
Until now Gaussian covariances have been used. However, a more general
model could be used in order to control the behaviour of the transiogram
at the origin. For instance, using the stable covariance model (Chiles and
Delfiner, 2012) with a geometrical anisotropy along stratigraphy and separable
anisotropy in the vertical direction would give transiograms of the form (for
one latent Gaussian function with a threshold at zero)
" " s !β # #
1 1 h2x h2y hz γ
t11 (hx , hy , hz ) = + arcsin exp − + 2 ∗exp[− ] cos[b hz ] .
2 π rx2 ry rz
(28)
The smoothness of facies boundaries decreases with the coefficients β and γ
(0 < β, γ ≤ 2).
Fig. 9 Three dimensional simulation with the cyclical PGS obtained with the parameters
shown in Fig. 8. The void represents the red facies in Fig. 8
with Φk random phases sampled from a uniform pdf between 0 and 2π and
<> the scalar product. The resulting field is Gaussian (central limit theorem)
when N tends to infinity. Based on the knowledge of x, the individual value
of Z(x) at each location x can be simulated, which enables the algorithm to
be coded in parallel and to be grid-free.
The spectral approach is well suited for a separable covariance as the multi-
dimensional Fourier transform of a separable covariance is the product of the
Fourier transforms. The three dimensional Fourier transform of a separable
covariance model ρ(hx , hy , hz ) (Eq. (25)) is
Fρ(hx ,hy ,hz ) (νz , νx , νy ) = Fρv (hz ) (νz ) Fρl (hx ,hy ) (νx , νy ), (30)
which means the frequencies νz in the vertical direction are sampled indepen-
dently from νx and νy according to the Fourier transform of their respective
covariance models.
Geostatistical modelling of cyclic and rhythmic facies architectures 17
In order to simulate the covariance model of Eq. (26), the Fourier trans-
forms of the covariance in the vertical direction and the lateral plane need to
be found (Eq. (30)). For the lateral plane, as a Gaussian covariance with geo-
metrical anisotropy is also separable, the two directions hx and hy are sampled
independently
Fρ(hx ,hy ) (νx , νy ) = Fρv (hx ) (νx ) Fρl (hy ) (νy ). (31)
" #
1 1 (b + νz )2 1 (b − νz )2
Fρz (hz ) (νz ) = √ exp(− ) + exp(− ) . (32)
2 rz π 4 rz2 4 rz2
Fig. 10 Variograms (grey) computed from ten realizations with model (black) and param-
eters: rx =0.15, rz =0.05, b = 100, N = 100. The simulated grid has 200 ∗ 200 cells of size
(0.005, 0.005)
with qi and qi0 being the thresholds associated with the truncation diagram.
The conditional simulation is usually performed in three steps (Chiles and
Delfiner, 2012). First, an unconditional simulation Z u (x) is performed as in
Sect. 4.1. Then, a local conditional simulation of Z(xi ) is carried out only
at data locations xi using Gibbs sampling such that Eq. (33) is respected
(Freulon and de Fouquet, 1993). Finally, simple kriging R(x) is calculated for
the difference at data locations between the conditional and unconditional
simulation
R(xi ) = Z(xi ) − Z u (xi ), (34)
is performed (Chiles and Delfiner, 2012)
X
R(x) = λi (x) R(xi ), (35)
i
with λi (x) the kriging weights (Chiles and Delfiner, 2012). Therefore, the con-
ditional Gaussian function Z at every location x is finally obtained by
The Gibbs sampling at data locations is not described here. Any algorithm
can be chosen (Emery et al, 2014) and applied with the desired covariance
model to simulate the values of Z(xi ) at the data locations. However, the
kriging of the difference (Eq. (35)) can be optimized when using a separable
covariance such as the model of Eq. (25). This is important because kriging
Geostatistical modelling of cyclic and rhythmic facies architectures 19
Fig. 11 The screening effect for simple kriging applied on two wells W1 and W2 to estimate
three locations on three different horizontal planes with one extended well (red)
over the whole domain is computationally expensive, and can make simulation
prohibitively slow. The simple kriging is optimized thanks to the well known
screening properties of separable covariances (Chiles and Delfiner, 2012). With
a separable anisotropy between the vertical direction and the lateral plane,
such as with the covariance model of Eq. (25), the weights λi (x) associated
with data located on different lateral planes from that of the estimated location
x are equal to zero (Chiles and Delfiner, 2012). This means that the estimation
of a location only depends on the data at the same horizontal level, and the
number of kriging weights therefore equals the number of wells intersecting this
horizontal level, as shown in Fig. 11. Assuming all wells are vertical and have
the same length, the number of weights for every kriged point is therefore the
number total of wells. If this is not the case, for instance because some vertical
wells do not penetrate a particular level, it is convenient to artificially extend
them by an unconditional simulation with Gibbs sampling (Fig. 11), so that
the geometrical configuration of the data points remains the same at all levels.
Therefore, the weigths are the same for every horizontal plane and the dual
form of two-dimensional kriging may be used, in which the data covariance
matrix is inverted only once (Chiles and Delfiner, 2012). This enables rapid
and efficient kriging.
In this paper, with the two latent Gaussian functions to condition (Eq. (21))
and the truncation rule of Fig. 2, the two conditions at each data point are
Fig. 12 Logs through the Upper Cyclic Facies interval of the Latemar carbonate platform
(Peterhänsel and Egenhoff, 2008) subtidal (red), intertidal (yellow) and supratidal (white)
subtidal (facies 1), intertidal (facies 2 and 3), supratidal storm deposits (facies
4 and 5). The next paragraph provides a quantitative analysis of the variations
of these three facies.
As illustrated by Fig. 12, the eight vertical logs show a high number of facies
transitions. The asymmetry is clear as the subtidal facies (red) tends to be
on top of the supratidal (white) facies. However, complete upward-shallowing
facies cycles, containing subtidal, intertidal and supratidal deposits, occur only
24 times, while there are 56 incomplete cycles in the eight logs, which means
that the sequences are pseudo-cyclic. The subtidal facies appears to show
regular spacing between beds within some wells (Fig. 12, in logs N8 and N16),
which would suggest a pseudo-rhythmicity of this facies. This is not the case
for the intertidal and supratidal facies, which show very different spacings
between the beds (Fig. 12).
All this information can be verified in the experimental transiograms com-
puted on the logs (Fig. 13, grey points). A dampened hole effect is observable
on the auto-transiogram of the subtidal deposits (Fig. 13a). The tangent at
the origin of the cross-transiogram of intertidal deposits overlain by subtidal
deposits (Fig. 13c) is low, showing that this transition is rare. More precisely,
T12 = 2.06 and T21 = 0.19 which means according to Eq. (7) that the intertidal
facies is four times more likely to overlie the subtidal facies than to underlie
it. Moreover, Pc ≈ 0.3, which clearly shows that the transitions between facies
are pseudo-cyclic.
Fig. 13 Experimental transiograms (grey points), model (black line) and realization (black
diamonds) after simulation with parameters in Table 2
gives an appropriate fit although the oscillations observed on the logs seem
less pronounced (Fig. 13a). This could be improved by using a more complex
covariance model and is discussed in Sect. 6.2. Then, the correlation coeffi-
cient ρ and the shift α can be chosen to fit the transition rates between the
subtidal and intertidal facies. Here the asymmetry (Eq. (7)) is clear, because
the probability of facies intertidal overlying facies subtidal is four times higher
that of the opposite. Therefore, the shift α is equal to the size of one verti-
cal cell and the correlation coefficient ρ is rather high (-0.6) which allows the
model to match these transition rates. The shift is strictly vertical because no
information on a possible lateral asymmetry is available (this topic is further
discussed in Sect. 6.3). Finally the covariance of the second Gaussian functions
ρ2 (h) determines the transiogram of intertidal facies. A rhythmic covariance is
not necessary here, as the transiogram does not show a clear hole effect. There
is not enough information to determine the lateral ranges of the covariance
directly from the data. Therefore, they are derived from visual comparison of
the realizations with the outcrop facies panel interpreted between the vertical
logs by Peterhänsel and Egenhoff (2008) and are chosen to be isotropic (in
the lateral plane). Finally, the thresholds q1 and q2 of the Gaussian functions
are determined according to the proportions of the facies and the correlation
coefficient ρ between the two Gaussian random functions (Armstrong et al,
2011). The parameters are summarized in Table 2.
Vertically, a Gaussian cosine covariance is chosen but laterally an exponen-
tial covariance (Chiles and Delfiner, 2012) is used as the Gaussian covariance
can cause numerical issues. A Gaussian or cubic covariance in the lateral di-
rection leads to a singular kriging system for the conditioning step as the
Geostatistical modelling of cyclic and rhythmic facies architectures 23
lateral range is large (Sect. 4.2). A stable covariance with a power close to two
could also be chosen to obtain a smoother model than that obtained with the
exponential.
The cell sizes of the grid are 0.1 meters vertically and 10 meters laterally
in both North and East directions, which gives approximately 800,000 cells.
The simulation takes approximately one minute to run with a standard Intel
processor i7, but it could be much faster by performing a parallel uncondi-
tional simulation with the spectral method, which is the longest computational
step. The conditioning by kriging every surface independently, as described in
Sect. 4.2, is almost instantaneous. One realization of the field is shown in
Fig. 14.
The visual aspect of the simulation is granular due to the exponential co-
variance model used in the lateral plane. Visually, the subtidal (red) facies
tends to lie on top of the supratidal (white) facies and rhythmicity is con-
firmed by the regular thickness between two subtidal (red) facies bodies. The
transiograms of the wells and the transiograms computed on one realization
are compared in Fig. 13. The sills derived from simulation (black diamonds in
Fig. 13) are accurate, which means that the facies proportions match those in
the wells. The tangent at the origin of the auto-transiograms is appropriately
fitted, which means the average thicknesses of different facies bodies honours
the well data. The tangent at the origin of the cross-transiograms matches
the transition rates of the wells (T12 = 2.1 and T21 = 0.19) which means the
asymmetry is respected. The hole effect in the realization transiogram is less
pronounced that the one used for the model, but is closer to the one observed
at the wells (grey circles in Fig. 13). This might be due to the condition-
ing to the wells, where the experimental transiograms show less pronounced
oscillations (black lines in Fig. 13).
6 Discussion
In order to illustrate the method, standard parameters have been chosen so far
for both the synthetic examples and the case study. However, some parameters
24 Thomas Le Blévec et al.
Fig. 14 One realization of the Latemar facies field from Peterhänsel and Egenhoff (2008)
data (Fig. 12), and two cross-sections obtained with the parameters summarized in Table 2
such as the truncation rule, the covariance model, or the shift can be made
more complex to adjust to different geological environments.
The same truncation rule has been applied through the paper (Fig. 2) with
the subtidal facies defined by the first Gaussian function and the two others
by both Gaussian functions. This has implications for the geometries of indi-
vidual facies bodies and for facies relationships. The facies defined by the first
Gaussian function erodes the two other facies (e.g. Fig. 8e), which means that
bodies of these facies can have very different geometries those of the other
facies. This behaviour can be reduced by increasing the correlation ρ between
the two Gaussian functions Z1 (x) and Z2 (x).
The truncation rule affects not only the facies transiograms but also higher
order statistical moments, which can have an impact on connectivity (Beucher
and Renard, 2016). However, these moments are not known analytically and so
are difficult to use for defining the truncation rule. It is recommended that the
earth modeller tries different truncation rules and inspects the visual aspect
of simulations so that it matches with his/her conceptual knowledge.
Geostatistical modelling of cyclic and rhythmic facies architectures 25
In this paper, only three facies have been used to illustrate the method. Two
methods for generalizing the truncation rule to more facies are found in the
literature: either the truncation is made more complex (Galli et al, 2006)
giving more deterministic relations between facies, or the number of Gaussian
functions is increased (Maleki et al, 2016). The first method would be too
limited to represent a large number of transition rates between facies, while
the second one should be able to model all transition rates (but the number of
parameters would be very high). The choice between the two methods should
depend on the case study and further work on this topic is required. It should
also be noted that some methods that create automatic truncation rules have
been developed (Deutsch and Deutsch, 2014; Astrakova et al, 2015). It would
be fruitful to generalize these methods by incorporating the shift between the
Gaussian functions in order to match asymmetric transition probabilities.
The vertical hole-effect model used in this paper is made of two parameters rz
and b (Eq. (23)) which provides some flexibility to match observed rhythmicity.
However, the case study shows that the observed transiograms can be even
more complex (Fig. 13) and two parameters might not be sufficient to represent
them. The covariance model could be modified to incorporate more than one
structure (Chiles and Delfiner, 2012). For instance a Gaussian covariance or
a cosine covariance can be added to the Gaussian cosine model (Eq. (23)).
The corresponding simulation can then be performed by summing together the
uncorrelated Gaussian random functions associated with each nested structure
of the covariance.
The method developed in this paper models cyclicity only in the vertical di-
rection, which is consistent with observations of most outcrop and subsur-
face data. However according to Walther’s law, the transitions between facies
should be equivalent laterally and vertically (Middleton, 1973). This means
that the facies ordering should be similar and the transition rates propor-
tional as in Markov Chains methods (Doveton, 1994; Purkis et al, 2012). Thus
vertical asymmetry should also be observed laterally. It is possible to model
such patterns with the presented method by defining a shift between the Gaus-
sian functions (Eq. (21)) with a lateral component, such that the asymmetry
is also lateral (Renard and Beucher, 2012; Le Blévec et al, 2016).
This choice will depend on the depositional environments, processes, con-
trols, and scale to be modelled. These aspects are typically interpreted with
reference to an underlying conceptual model, such as those for allocyclically
26 Thomas Le Blévec et al.
7 Conclusion
While cyclicity and rhythmicity are commonly observed in geological data sets,
few existing geostatistical algorithm can model both patterns in an efficient
manner. By addressing this issue, the method developed here is promising for
modelling carbonate or clastic reservoirs that contain such cyclical facies suc-
cessions. Broadly speaking, cyclicity and rhythmicity are quantified by facies
transiograms that are computed from data (e.g. vertical facies successions)
and fitted with an advanced truncated Pluri-Gaussian model for performing
three dimensional simulations.
The model used for the latent Gaussian functions is the linear model of
coregionalization with a spatial shift, which creates the asymmetric cycles.
The covariance of the Gaussian functions presents a dampened hole effect for
capturing the rhythmicity. As this hole effect is generally observed only in
the vertical direction, a separable covariance model, which is the product of
a lateral and a vertical covariance is used so that no rhythmicity is modelled
laterally along the stratigraphy. The space-time separable covariance is simu-
lated readily by the continuous spectral method. The numerical properties of
separable covariance allows rapid and efficient conditioning to data via krig-
ing of every horizontal surface independently. The procedure has been applied
successfully to model a carbonate platform environment that shows cyclicity
and rhythmicity in facies architectures.
Acknowledgements The authors would like to thank Total for funding Olivier Dubrule
professorship at Imperial College and the Department of Earth Science and Engineering at
Imperial College for a scholarship to Thomas Le Blevec.
Geostatistical modelling of cyclic and rhythmic facies architectures 27
References